Re: agent crash when proxying for Squid

[John H. <jh...@ve...>]

Wes Hardaker wrote:

>>>>>>On Mon, 02 May 2005 13:12:47 -0700, John Hardin <jh...@ve...> said:
>>>>>>            
>>>>>>
>
>John> Apparently the BULK-GET received by snmpd is sent to Squid as
>John> a GET-NEXT with two OIDs.  Ethereal reports this:
>
>John> Version: 1 (0)
>John> Community: public
>John> PDU type: GET-NEXT (1)
>John> Request Id: 0x60962325
>John> Error Status: NO ERROR (0)
>John> Error Index: 0
>John> Object identifier 1: 1.3.6.1.4.1.3495.1.1.1 
>John> (SNMPv2-SMI::enterprises.3495.1.1.1)
>John> Value: NULL
>John> Object identifier 2: 1.3.6.1.4.1.3495.1.1.2 
>John> (SNMPv2-SMI::enterprises.3495.1.1.2)
>John> Value: NULL
>
>
>John> Squid responds with this:
>
>John> Version: 1 (0)
>John> Community: public
>John> PDU type: RESPONSE (2)
>John> Request Id: 0x60962325
>John> Error Status: NO ERROR (0)
>John> Error Index: 0
>John> Object identifier 1: 1.3.6.1.4.1.3495.1.1.1.0 
>John> (SNMPv2-SMI::enterprises.3495.1.1.1.0)
>John> Value: INTEGER: 84
>
>John> It looks like the agent then segfaults when the snmpbulkget
>John> retries.
>
>hmmm...  lame that it's only returning one varbind when 2 were
>requested!  not that anything should crash when this happens, mind you...
>
>  
>
Yes, lame.  I've found the code in Squid that responds to SNMP requests and,
sure enough, if it's a GET it will loop through all the OIDs but if it's 
a GETNEXT
it just assumes a single OID.  If I get the fix working I'll submit it 
the the Squid
folks.